moab::anonymous_namespace{IntegerReprosum.cpp} Namespace Reference

Functions
double	frac_and_exp (double x, int &exp_out)
	Fortran-equivalent fraction(x) for radix-2 doubles: returns f in [0.5, 1) (or in (-1, -0.5] for negative x) and stores the exponent so that x = f * 2^e Same as std::frexp. More...
double	scale2 (double x, int n)
	Fortran-equivalent scale(x, n) for radix-2 doubles: returns x * 2^n. Same as std::ldexp. More...
double	set_exp (double x, int e)
	Fortran-equivalent set_exponent(x, e): returns a value with the same fraction as x but with exponent e. Same as ldexp(frexp(x, &dummy), e) More...
int64_t	i2pow (int k)
	MPI_INT64_T equivalent for our int64 type. Use MPI_LONG_LONG_INT for portability across MPI implementations that don't yet expose MPI_INT64_T as a primary datatype. More...

Function Documentation

frac_and_exp()

double moab::anonymous_namespace{IntegerReprosum.cpp}::frac_and_exp ( double  x, int &  exp_out  )

inline

Fortran-equivalent fraction(x) for radix-2 doubles: returns f in [0.5, 1) (or in (-1, -0.5] for negative x) and stores the exponent so that x = f * 2^e Same as std::frexp.

Definition at line 49 of file IntegerReprosum.cpp.

{
    return std::frexp( x, &exp_out );
}

i2pow()

int64_t moab::anonymous_namespace{IntegerReprosum.cpp}::i2pow ( int  k )

inline

MPI_INT64_T equivalent for our int64 type. Use MPI_LONG_LONG_INT for portability across MPI implementations that don't yet expose MPI_INT64_T as a primary datatype.

Integer power 2^k (for small k <= 62).

Definition at line 86 of file IntegerReprosum.cpp.

{
    return ( static_cast< int64_t >( 1 ) << k );
}

Referenced by moab::IntegerReprosum::decode_global(), and moab::IntegerReprosum::encode_local().

scale2()

double moab::anonymous_namespace{IntegerReprosum.cpp}::scale2 ( double  x, int  n  )

inline

Fortran-equivalent scale(x, n) for radix-2 doubles: returns x * 2^n. Same as std::ldexp.

Definition at line 56 of file IntegerReprosum.cpp.

{
    return std::ldexp( x, n );
}

Referenced by moab::IntegerReprosum::decode_global(), and moab::IntegerReprosum::encode_local().

set_exp()

double moab::anonymous_namespace{IntegerReprosum.cpp}::set_exp ( double  x, int  e  )

inline

Fortran-equivalent set_exponent(x, e): returns a value with the same fraction as x but with exponent e. Same as ldexp(frexp(x, &dummy), e)

Definition at line 64 of file IntegerReprosum.cpp.

{
    int dummy;
    double f = std::frexp( x, &dummy );
    return std::ldexp( f, e );
}

Referenced by moab::IntegerReprosum::decode_global().